{ "best_metric": null, "best_model_checkpoint": null, "epoch": 49.99935691318328, "global_step": 38850, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.13, "learning_rate": 3.675e-06, "loss": 21.0125, "step": 100 }, { "epoch": 0.26, "learning_rate": 7.425e-06, "loss": 10.9603, "step": 200 }, { "epoch": 0.39, "learning_rate": 1.1174999999999999e-05, "loss": 6.2714, "step": 300 }, { "epoch": 0.51, "learning_rate": 1.4925e-05, "loss": 5.0154, "step": 400 }, { "epoch": 0.64, "learning_rate": 1.8675e-05, "loss": 4.1367, "step": 500 }, { "epoch": 0.64, "eval_loss": 3.88252592086792, "eval_runtime": 666.8464, "eval_samples_per_second": 12.448, "eval_steps_per_second": 1.557, "eval_wer": 1.0, "step": 500 }, { "epoch": 0.77, "learning_rate": 2.2424999999999996e-05, "loss": 3.5318, "step": 600 }, { "epoch": 0.9, "learning_rate": 2.6174999999999996e-05, "loss": 3.2175, "step": 700 }, { "epoch": 1.03, "learning_rate": 2.9925e-05, "loss": 3.1431, "step": 800 }, { "epoch": 1.16, "learning_rate": 3.3675e-05, "loss": 3.1247, "step": 900 }, { "epoch": 1.29, "learning_rate": 3.7424999999999995e-05, "loss": 2.9677, "step": 1000 }, { "epoch": 1.29, "eval_loss": 2.949765205383301, "eval_runtime": 679.306, "eval_samples_per_second": 12.22, "eval_steps_per_second": 1.528, "eval_wer": 1.0, "step": 1000 }, { "epoch": 1.42, "learning_rate": 4.1175e-05, "loss": 2.9224, "step": 1100 }, { "epoch": 1.54, "learning_rate": 4.4924999999999994e-05, "loss": 2.8662, "step": 1200 }, { "epoch": 1.67, "learning_rate": 4.8675e-05, "loss": 2.591, "step": 1300 }, { "epoch": 1.8, "learning_rate": 5.2424999999999994e-05, "loss": 1.959, "step": 1400 }, { "epoch": 1.93, "learning_rate": 5.6175e-05, "loss": 1.5884, "step": 1500 }, { "epoch": 1.93, "eval_loss": 0.6722400188446045, "eval_runtime": 671.1611, "eval_samples_per_second": 12.368, "eval_steps_per_second": 1.547, "eval_wer": 0.6492582632081201, "step": 1500 }, { "epoch": 2.06, "learning_rate": 5.9925e-05, "loss": 1.4568, "step": 1600 }, { "epoch": 2.19, "learning_rate": 6.367499999999999e-05, "loss": 1.3268, "step": 1700 }, { "epoch": 2.32, "learning_rate": 6.7425e-05, "loss": 1.29, "step": 1800 }, { "epoch": 2.45, "learning_rate": 7.1175e-05, "loss": 1.2493, "step": 1900 }, { "epoch": 2.57, "learning_rate": 7.492499999999999e-05, "loss": 1.2292, "step": 2000 }, { "epoch": 2.57, "eval_loss": 0.3634721636772156, "eval_runtime": 665.3111, "eval_samples_per_second": 12.477, "eval_steps_per_second": 1.56, "eval_wer": 0.3202047367051271, "step": 2000 }, { "epoch": 2.7, "learning_rate": 7.480054274084124e-05, "loss": 1.1888, "step": 2100 }, { "epoch": 2.83, "learning_rate": 7.459905020352781e-05, "loss": 1.168, "step": 2200 }, { "epoch": 2.96, "learning_rate": 7.43955223880597e-05, "loss": 1.1837, "step": 2300 }, { "epoch": 3.09, "learning_rate": 7.419199457259158e-05, "loss": 1.1336, "step": 2400 }, { "epoch": 3.22, "learning_rate": 7.399050203527815e-05, "loss": 1.1314, "step": 2500 }, { "epoch": 3.22, "eval_loss": 0.2969689965248108, "eval_runtime": 670.1792, "eval_samples_per_second": 12.386, "eval_steps_per_second": 1.549, "eval_wer": 0.2680142274659495, "step": 2500 }, { "epoch": 3.35, "learning_rate": 7.378697421981004e-05, "loss": 1.1159, "step": 2600 }, { "epoch": 3.47, "learning_rate": 7.358344640434192e-05, "loss": 1.0976, "step": 2700 }, { "epoch": 3.6, "learning_rate": 7.33799185888738e-05, "loss": 1.102, "step": 2800 }, { "epoch": 3.73, "learning_rate": 7.317639077340569e-05, "loss": 1.0915, "step": 2900 }, { "epoch": 3.86, "learning_rate": 7.297286295793758e-05, "loss": 1.0879, "step": 3000 }, { "epoch": 3.86, "eval_loss": 0.2671239972114563, "eval_runtime": 666.0789, "eval_samples_per_second": 12.462, "eval_steps_per_second": 1.558, "eval_wer": 0.24856424047887568, "step": 3000 }, { "epoch": 3.99, "learning_rate": 7.276933514246946e-05, "loss": 1.0838, "step": 3100 }, { "epoch": 4.12, "learning_rate": 7.256580732700136e-05, "loss": 1.0662, "step": 3200 }, { "epoch": 4.25, "learning_rate": 7.236227951153323e-05, "loss": 1.0418, "step": 3300 }, { "epoch": 4.38, "learning_rate": 7.215875169606512e-05, "loss": 1.0358, "step": 3400 }, { "epoch": 4.5, "learning_rate": 7.19572591587517e-05, "loss": 1.0344, "step": 3500 }, { "epoch": 4.5, "eval_loss": 0.26251843571662903, "eval_runtime": 663.0388, "eval_samples_per_second": 12.52, "eval_steps_per_second": 1.566, "eval_wer": 0.2239264335907001, "step": 3500 }, { "epoch": 4.63, "learning_rate": 7.175373134328358e-05, "loss": 1.0395, "step": 3600 }, { "epoch": 4.76, "learning_rate": 7.155020352781546e-05, "loss": 1.031, "step": 3700 }, { "epoch": 4.89, "learning_rate": 7.134667571234734e-05, "loss": 1.016, "step": 3800 }, { "epoch": 5.02, "learning_rate": 7.114314789687924e-05, "loss": 1.0319, "step": 3900 }, { "epoch": 5.15, "learning_rate": 7.093962008141112e-05, "loss": 1.0109, "step": 4000 }, { "epoch": 5.15, "eval_loss": 0.2520427405834198, "eval_runtime": 660.1809, "eval_samples_per_second": 12.574, "eval_steps_per_second": 1.572, "eval_wer": 0.22297215233798906, "step": 4000 }, { "epoch": 5.28, "learning_rate": 7.073609226594301e-05, "loss": 1.0109, "step": 4100 }, { "epoch": 5.41, "learning_rate": 7.05325644504749e-05, "loss": 1.0064, "step": 4200 }, { "epoch": 5.53, "learning_rate": 7.032903663500678e-05, "loss": 0.9985, "step": 4300 }, { "epoch": 5.66, "learning_rate": 7.012550881953866e-05, "loss": 1.0407, "step": 4400 }, { "epoch": 5.79, "learning_rate": 6.992198100407055e-05, "loss": 0.9966, "step": 4500 }, { "epoch": 5.79, "eval_loss": 0.22795377671718597, "eval_runtime": 658.6571, "eval_samples_per_second": 12.603, "eval_steps_per_second": 1.576, "eval_wer": 0.2105144443480524, "step": 4500 }, { "epoch": 5.92, "learning_rate": 6.971845318860243e-05, "loss": 0.9938, "step": 4600 }, { "epoch": 6.05, "learning_rate": 6.951492537313432e-05, "loss": 1.007, "step": 4700 }, { "epoch": 6.18, "learning_rate": 6.93113975576662e-05, "loss": 0.9835, "step": 4800 }, { "epoch": 6.31, "learning_rate": 6.910786974219809e-05, "loss": 0.9927, "step": 4900 }, { "epoch": 6.43, "learning_rate": 6.890434192672997e-05, "loss": 0.9815, "step": 5000 }, { "epoch": 6.43, "eval_loss": 0.225350484251976, "eval_runtime": 716.6616, "eval_samples_per_second": 11.583, "eval_steps_per_second": 1.448, "eval_wer": 0.2179057864145051, "step": 5000 }, { "epoch": 6.56, "learning_rate": 6.870081411126187e-05, "loss": 0.9874, "step": 5100 }, { "epoch": 6.69, "learning_rate": 6.849728629579376e-05, "loss": 0.9856, "step": 5200 }, { "epoch": 6.82, "learning_rate": 6.829375848032564e-05, "loss": 0.9915, "step": 5300 }, { "epoch": 6.95, "learning_rate": 6.809023066485753e-05, "loss": 1.0046, "step": 5400 }, { "epoch": 7.08, "learning_rate": 6.78887381275441e-05, "loss": 0.9744, "step": 5500 }, { "epoch": 7.08, "eval_loss": 0.23012416064739227, "eval_runtime": 684.7726, "eval_samples_per_second": 12.122, "eval_steps_per_second": 1.516, "eval_wer": 0.21365489719788322, "step": 5500 }, { "epoch": 7.21, "learning_rate": 6.768724559023065e-05, "loss": 0.9836, "step": 5600 }, { "epoch": 7.34, "learning_rate": 6.748371777476255e-05, "loss": 0.9752, "step": 5700 }, { "epoch": 7.46, "learning_rate": 6.728018995929444e-05, "loss": 0.9677, "step": 5800 }, { "epoch": 7.59, "learning_rate": 6.707666214382632e-05, "loss": 0.9678, "step": 5900 }, { "epoch": 7.72, "learning_rate": 6.68731343283582e-05, "loss": 0.9487, "step": 6000 }, { "epoch": 7.72, "eval_loss": 0.22240765392780304, "eval_runtime": 695.5316, "eval_samples_per_second": 11.935, "eval_steps_per_second": 1.492, "eval_wer": 0.20506636592348398, "step": 6000 }, { "epoch": 7.85, "learning_rate": 6.666960651289009e-05, "loss": 0.9644, "step": 6100 }, { "epoch": 7.98, "learning_rate": 6.646607869742198e-05, "loss": 0.9797, "step": 6200 }, { "epoch": 8.11, "learning_rate": 6.626255088195386e-05, "loss": 0.9697, "step": 6300 }, { "epoch": 8.24, "learning_rate": 6.605902306648575e-05, "loss": 0.9502, "step": 6400 }, { "epoch": 8.37, "learning_rate": 6.585549525101763e-05, "loss": 0.9431, "step": 6500 }, { "epoch": 8.37, "eval_loss": 0.21053080260753632, "eval_runtime": 691.0445, "eval_samples_per_second": 12.012, "eval_steps_per_second": 1.502, "eval_wer": 0.19918452329313785, "step": 6500 }, { "epoch": 8.49, "learning_rate": 6.565196743554952e-05, "loss": 0.9414, "step": 6600 }, { "epoch": 8.62, "learning_rate": 6.54484396200814e-05, "loss": 0.9491, "step": 6700 }, { "epoch": 8.75, "learning_rate": 6.524491180461329e-05, "loss": 0.9548, "step": 6800 }, { "epoch": 8.88, "learning_rate": 6.504138398914518e-05, "loss": 0.9409, "step": 6900 }, { "epoch": 9.01, "learning_rate": 6.483785617367707e-05, "loss": 0.9365, "step": 7000 }, { "epoch": 9.01, "eval_loss": 0.21139565110206604, "eval_runtime": 668.2218, "eval_samples_per_second": 12.423, "eval_steps_per_second": 1.553, "eval_wer": 0.20185651080072872, "step": 7000 }, { "epoch": 9.14, "learning_rate": 6.463432835820895e-05, "loss": 0.9264, "step": 7100 }, { "epoch": 9.27, "learning_rate": 6.443080054274083e-05, "loss": 0.9129, "step": 7200 }, { "epoch": 9.39, "learning_rate": 6.422727272727272e-05, "loss": 0.9267, "step": 7300 }, { "epoch": 9.52, "learning_rate": 6.402374491180461e-05, "loss": 0.9213, "step": 7400 }, { "epoch": 9.65, "learning_rate": 6.38202170963365e-05, "loss": 0.9268, "step": 7500 }, { "epoch": 9.65, "eval_loss": 0.20965604484081268, "eval_runtime": 672.8441, "eval_samples_per_second": 12.337, "eval_steps_per_second": 1.543, "eval_wer": 0.19880281079205345, "step": 7500 }, { "epoch": 9.78, "learning_rate": 6.361668928086838e-05, "loss": 0.9087, "step": 7600 }, { "epoch": 9.91, "learning_rate": 6.341316146540027e-05, "loss": 0.9142, "step": 7700 }, { "epoch": 10.04, "learning_rate": 6.320963364993215e-05, "loss": 0.9274, "step": 7800 }, { "epoch": 10.17, "learning_rate": 6.300610583446404e-05, "loss": 0.9223, "step": 7900 }, { "epoch": 10.3, "learning_rate": 6.280257801899593e-05, "loss": 0.9292, "step": 8000 }, { "epoch": 10.3, "eval_loss": 0.21196416020393372, "eval_runtime": 663.5835, "eval_samples_per_second": 12.509, "eval_steps_per_second": 1.564, "eval_wer": 0.1985599028368179, "step": 8000 }, { "epoch": 10.42, "learning_rate": 6.25990502035278e-05, "loss": 0.9355, "step": 8100 }, { "epoch": 10.55, "learning_rate": 6.239552238805969e-05, "loss": 0.9095, "step": 8200 }, { "epoch": 10.68, "learning_rate": 6.219199457259158e-05, "loss": 0.8979, "step": 8300 }, { "epoch": 10.81, "learning_rate": 6.198846675712346e-05, "loss": 0.9174, "step": 8400 }, { "epoch": 10.94, "learning_rate": 6.178493894165536e-05, "loss": 0.929, "step": 8500 }, { "epoch": 10.94, "eval_loss": 0.2047712206840515, "eval_runtime": 660.2531, "eval_samples_per_second": 12.572, "eval_steps_per_second": 1.572, "eval_wer": 0.1998264943176889, "step": 8500 }, { "epoch": 11.07, "learning_rate": 6.158141112618724e-05, "loss": 0.921, "step": 8600 }, { "epoch": 11.2, "learning_rate": 6.137788331071913e-05, "loss": 0.9058, "step": 8700 }, { "epoch": 11.33, "learning_rate": 6.117435549525101e-05, "loss": 0.887, "step": 8800 }, { "epoch": 11.45, "learning_rate": 6.09708276797829e-05, "loss": 0.9142, "step": 8900 }, { "epoch": 11.58, "learning_rate": 6.076729986431478e-05, "loss": 0.9017, "step": 9000 }, { "epoch": 11.58, "eval_loss": 0.20354785025119781, "eval_runtime": 663.071, "eval_samples_per_second": 12.519, "eval_steps_per_second": 1.565, "eval_wer": 0.19993059772707555, "step": 9000 }, { "epoch": 11.71, "learning_rate": 6.056377204884667e-05, "loss": 0.9031, "step": 9100 }, { "epoch": 11.84, "learning_rate": 6.0360244233378555e-05, "loss": 0.8996, "step": 9200 }, { "epoch": 11.97, "learning_rate": 6.015671641791044e-05, "loss": 0.9138, "step": 9300 }, { "epoch": 12.1, "learning_rate": 5.9953188602442325e-05, "loss": 0.8957, "step": 9400 }, { "epoch": 12.23, "learning_rate": 5.974966078697422e-05, "loss": 0.8898, "step": 9500 }, { "epoch": 12.23, "eval_loss": 0.1960887461900711, "eval_runtime": 671.0322, "eval_samples_per_second": 12.37, "eval_steps_per_second": 1.547, "eval_wer": 0.1907521471328186, "step": 9500 }, { "epoch": 12.35, "learning_rate": 5.95461329715061e-05, "loss": 0.884, "step": 9600 }, { "epoch": 12.48, "learning_rate": 5.934260515603799e-05, "loss": 0.8872, "step": 9700 }, { "epoch": 12.61, "learning_rate": 5.913907734056988e-05, "loss": 0.8993, "step": 9800 }, { "epoch": 12.74, "learning_rate": 5.893554952510176e-05, "loss": 0.8845, "step": 9900 }, { "epoch": 12.87, "learning_rate": 5.873202170963364e-05, "loss": 0.8799, "step": 10000 }, { "epoch": 12.87, "eval_loss": 0.19447088241577148, "eval_runtime": 677.1371, "eval_samples_per_second": 12.259, "eval_steps_per_second": 1.533, "eval_wer": 0.1816951505161794, "step": 10000 }, { "epoch": 13.0, "learning_rate": 5.852849389416553e-05, "loss": 0.8997, "step": 10100 }, { "epoch": 13.13, "learning_rate": 5.832496607869741e-05, "loss": 0.8774, "step": 10200 }, { "epoch": 13.26, "learning_rate": 5.8121438263229305e-05, "loss": 0.8761, "step": 10300 }, { "epoch": 13.38, "learning_rate": 5.791791044776119e-05, "loss": 0.8731, "step": 10400 }, { "epoch": 13.51, "learning_rate": 5.7714382632293075e-05, "loss": 0.869, "step": 10500 }, { "epoch": 13.51, "eval_loss": 0.19293242692947388, "eval_runtime": 658.1115, "eval_samples_per_second": 12.613, "eval_steps_per_second": 1.577, "eval_wer": 0.18436713802377028, "step": 10500 }, { "epoch": 13.64, "learning_rate": 5.751085481682497e-05, "loss": 0.8771, "step": 10600 }, { "epoch": 13.77, "learning_rate": 5.730936227951153e-05, "loss": 0.8718, "step": 10700 }, { "epoch": 13.9, "learning_rate": 5.710583446404341e-05, "loss": 0.8873, "step": 10800 }, { "epoch": 14.03, "learning_rate": 5.6902306648575305e-05, "loss": 0.878, "step": 10900 }, { "epoch": 14.16, "learning_rate": 5.669877883310719e-05, "loss": 0.8572, "step": 11000 }, { "epoch": 14.16, "eval_loss": 0.19409486651420593, "eval_runtime": 670.5041, "eval_samples_per_second": 12.38, "eval_steps_per_second": 1.548, "eval_wer": 0.18882623405916543, "step": 11000 }, { "epoch": 14.29, "learning_rate": 5.6495251017639075e-05, "loss": 0.8561, "step": 11100 }, { "epoch": 14.41, "learning_rate": 5.6291723202170953e-05, "loss": 0.8739, "step": 11200 }, { "epoch": 14.54, "learning_rate": 5.608819538670284e-05, "loss": 0.8706, "step": 11300 }, { "epoch": 14.67, "learning_rate": 5.588466757123473e-05, "loss": 0.8693, "step": 11400 }, { "epoch": 14.8, "learning_rate": 5.5681139755766616e-05, "loss": 0.8691, "step": 11500 }, { "epoch": 14.8, "eval_loss": 0.1912171095609665, "eval_runtime": 691.3958, "eval_samples_per_second": 12.006, "eval_steps_per_second": 1.501, "eval_wer": 0.18037650733061508, "step": 11500 }, { "epoch": 14.93, "learning_rate": 5.54776119402985e-05, "loss": 0.8562, "step": 11600 }, { "epoch": 15.06, "learning_rate": 5.527408412483039e-05, "loss": 0.8874, "step": 11700 }, { "epoch": 15.19, "learning_rate": 5.507055630936228e-05, "loss": 0.8503, "step": 11800 }, { "epoch": 15.32, "learning_rate": 5.486702849389416e-05, "loss": 0.8558, "step": 11900 }, { "epoch": 15.44, "learning_rate": 5.466350067842605e-05, "loss": 0.8645, "step": 12000 }, { "epoch": 15.44, "eval_loss": 0.1950196474790573, "eval_runtime": 670.3998, "eval_samples_per_second": 12.382, "eval_steps_per_second": 1.548, "eval_wer": 0.18506116075301465, "step": 12000 }, { "epoch": 15.57, "learning_rate": 5.4459972862957926e-05, "loss": 0.8668, "step": 12100 }, { "epoch": 15.7, "learning_rate": 5.425644504748982e-05, "loss": 0.8461, "step": 12200 }, { "epoch": 15.83, "learning_rate": 5.40529172320217e-05, "loss": 0.8508, "step": 12300 }, { "epoch": 15.96, "learning_rate": 5.384938941655359e-05, "loss": 0.8529, "step": 12400 }, { "epoch": 16.09, "learning_rate": 5.364586160108548e-05, "loss": 0.8468, "step": 12500 }, { "epoch": 16.09, "eval_loss": 0.1879136562347412, "eval_runtime": 672.4666, "eval_samples_per_second": 12.344, "eval_steps_per_second": 1.544, "eval_wer": 0.17701049709377983, "step": 12500 }, { "epoch": 16.22, "learning_rate": 5.3442333785617365e-05, "loss": 0.8342, "step": 12600 }, { "epoch": 16.34, "learning_rate": 5.323880597014925e-05, "loss": 0.8465, "step": 12700 }, { "epoch": 16.47, "learning_rate": 5.303731343283582e-05, "loss": 0.8206, "step": 12800 }, { "epoch": 16.6, "learning_rate": 5.28337856173677e-05, "loss": 0.8453, "step": 12900 }, { "epoch": 16.73, "learning_rate": 5.263025780189959e-05, "loss": 0.8405, "step": 13000 }, { "epoch": 16.73, "eval_loss": 0.18807044625282288, "eval_runtime": 680.7317, "eval_samples_per_second": 12.194, "eval_steps_per_second": 1.525, "eval_wer": 0.1758827101587577, "step": 13000 }, { "epoch": 16.86, "learning_rate": 5.2426729986431473e-05, "loss": 0.8571, "step": 13100 }, { "epoch": 16.99, "learning_rate": 5.2223202170963365e-05, "loss": 0.854, "step": 13200 }, { "epoch": 17.12, "learning_rate": 5.201967435549525e-05, "loss": 0.8578, "step": 13300 }, { "epoch": 17.25, "learning_rate": 5.181614654002713e-05, "loss": 0.8382, "step": 13400 }, { "epoch": 17.37, "learning_rate": 5.1612618724559014e-05, "loss": 0.8647, "step": 13500 }, { "epoch": 17.37, "eval_loss": 0.1861347109079361, "eval_runtime": 674.9378, "eval_samples_per_second": 12.299, "eval_steps_per_second": 1.538, "eval_wer": 0.17395679708510453, "step": 13500 }, { "epoch": 17.5, "learning_rate": 5.1409090909090906e-05, "loss": 0.8372, "step": 13600 }, { "epoch": 17.63, "learning_rate": 5.120556309362279e-05, "loss": 0.8253, "step": 13700 }, { "epoch": 17.76, "learning_rate": 5.1002035278154676e-05, "loss": 0.8353, "step": 13800 }, { "epoch": 17.89, "learning_rate": 5.079850746268656e-05, "loss": 0.8548, "step": 13900 }, { "epoch": 18.02, "learning_rate": 5.059497964721845e-05, "loss": 0.8477, "step": 14000 }, { "epoch": 18.02, "eval_loss": 0.1782081127166748, "eval_runtime": 701.0008, "eval_samples_per_second": 11.842, "eval_steps_per_second": 1.481, "eval_wer": 0.17017437321072265, "step": 14000 }, { "epoch": 18.15, "learning_rate": 5.039145183175034e-05, "loss": 0.8121, "step": 14100 }, { "epoch": 18.28, "learning_rate": 5.018792401628222e-05, "loss": 0.8243, "step": 14200 }, { "epoch": 18.4, "learning_rate": 4.99843962008141e-05, "loss": 0.8312, "step": 14300 }, { "epoch": 18.53, "learning_rate": 4.978086838534599e-05, "loss": 0.8288, "step": 14400 }, { "epoch": 18.66, "learning_rate": 4.957734056987788e-05, "loss": 0.811, "step": 14500 }, { "epoch": 18.66, "eval_loss": 0.19152408838272095, "eval_runtime": 690.1271, "eval_samples_per_second": 12.028, "eval_steps_per_second": 1.504, "eval_wer": 0.1757265550446777, "step": 14500 }, { "epoch": 18.79, "learning_rate": 4.937381275440976e-05, "loss": 0.8276, "step": 14600 }, { "epoch": 18.92, "learning_rate": 4.917028493894165e-05, "loss": 0.8337, "step": 14700 }, { "epoch": 19.05, "learning_rate": 4.896675712347354e-05, "loss": 0.8343, "step": 14800 }, { "epoch": 19.18, "learning_rate": 4.8763229308005425e-05, "loss": 0.8046, "step": 14900 }, { "epoch": 19.3, "learning_rate": 4.855970149253731e-05, "loss": 0.8165, "step": 15000 }, { "epoch": 19.3, "eval_loss": 0.18196436762809753, "eval_runtime": 694.7382, "eval_samples_per_second": 11.948, "eval_steps_per_second": 1.494, "eval_wer": 0.172447297648998, "step": 15000 }, { "epoch": 19.43, "learning_rate": 4.8356173677069196e-05, "loss": 0.8109, "step": 15100 }, { "epoch": 19.56, "learning_rate": 4.815264586160108e-05, "loss": 0.824, "step": 15200 }, { "epoch": 19.69, "learning_rate": 4.7949118046132966e-05, "loss": 0.8213, "step": 15300 }, { "epoch": 19.82, "learning_rate": 4.774559023066485e-05, "loss": 0.8197, "step": 15400 }, { "epoch": 19.95, "learning_rate": 4.7542062415196736e-05, "loss": 0.8166, "step": 15500 }, { "epoch": 19.95, "eval_loss": 0.17982707917690277, "eval_runtime": 678.5205, "eval_samples_per_second": 12.234, "eval_steps_per_second": 1.53, "eval_wer": 0.16974060900494492, "step": 15500 }, { "epoch": 20.08, "learning_rate": 4.733853459972863e-05, "loss": 0.8024, "step": 15600 }, { "epoch": 20.21, "learning_rate": 4.713500678426051e-05, "loss": 0.8093, "step": 15700 }, { "epoch": 20.33, "learning_rate": 4.69314789687924e-05, "loss": 0.8215, "step": 15800 }, { "epoch": 20.46, "learning_rate": 4.672795115332428e-05, "loss": 0.8061, "step": 15900 }, { "epoch": 20.59, "learning_rate": 4.652645861601085e-05, "loss": 0.8167, "step": 16000 }, { "epoch": 20.59, "eval_loss": 0.18053852021694183, "eval_runtime": 661.7443, "eval_samples_per_second": 12.544, "eval_steps_per_second": 1.569, "eval_wer": 0.1751539862930511, "step": 16000 }, { "epoch": 20.72, "learning_rate": 4.6322930800542736e-05, "loss": 0.8003, "step": 16100 }, { "epoch": 20.85, "learning_rate": 4.6121438263229304e-05, "loss": 0.8008, "step": 16200 }, { "epoch": 20.98, "learning_rate": 4.591791044776119e-05, "loss": 0.8064, "step": 16300 }, { "epoch": 21.11, "learning_rate": 4.5714382632293075e-05, "loss": 0.8184, "step": 16400 }, { "epoch": 21.24, "learning_rate": 4.551085481682496e-05, "loss": 0.7908, "step": 16500 }, { "epoch": 21.24, "eval_loss": 0.17608480155467987, "eval_runtime": 665.8329, "eval_samples_per_second": 12.467, "eval_steps_per_second": 1.559, "eval_wer": 0.169914114687256, "step": 16500 }, { "epoch": 21.36, "learning_rate": 4.530732700135685e-05, "loss": 0.7861, "step": 16600 }, { "epoch": 21.49, "learning_rate": 4.5103799185888737e-05, "loss": 0.8065, "step": 16700 }, { "epoch": 21.62, "learning_rate": 4.490027137042062e-05, "loss": 0.7982, "step": 16800 }, { "epoch": 21.75, "learning_rate": 4.46967435549525e-05, "loss": 0.803, "step": 16900 }, { "epoch": 21.88, "learning_rate": 4.449321573948439e-05, "loss": 0.7925, "step": 17000 }, { "epoch": 21.88, "eval_loss": 0.1740112453699112, "eval_runtime": 660.0442, "eval_samples_per_second": 12.576, "eval_steps_per_second": 1.573, "eval_wer": 0.17093779821289148, "step": 17000 }, { "epoch": 22.01, "learning_rate": 4.428968792401628e-05, "loss": 0.7822, "step": 17100 }, { "epoch": 22.14, "learning_rate": 4.408616010854816e-05, "loss": 0.7851, "step": 17200 }, { "epoch": 22.26, "learning_rate": 4.388263229308005e-05, "loss": 0.7871, "step": 17300 }, { "epoch": 22.39, "learning_rate": 4.367910447761194e-05, "loss": 0.7964, "step": 17400 }, { "epoch": 22.52, "learning_rate": 4.3475576662143824e-05, "loss": 0.7803, "step": 17500 }, { "epoch": 22.52, "eval_loss": 0.1815224587917328, "eval_runtime": 660.2602, "eval_samples_per_second": 12.572, "eval_steps_per_second": 1.572, "eval_wer": 0.17269020560423354, "step": 17500 }, { "epoch": 22.65, "learning_rate": 4.327204884667571e-05, "loss": 0.7793, "step": 17600 }, { "epoch": 22.78, "learning_rate": 4.30685210312076e-05, "loss": 0.7872, "step": 17700 }, { "epoch": 22.91, "learning_rate": 4.286499321573948e-05, "loss": 0.7895, "step": 17800 }, { "epoch": 23.04, "learning_rate": 4.2661465400271364e-05, "loss": 0.7925, "step": 17900 }, { "epoch": 23.17, "learning_rate": 4.245793758480325e-05, "loss": 0.7839, "step": 18000 }, { "epoch": 23.17, "eval_loss": 0.17369158565998077, "eval_runtime": 664.004, "eval_samples_per_second": 12.501, "eval_steps_per_second": 1.563, "eval_wer": 0.1693588965038605, "step": 18000 }, { "epoch": 23.29, "learning_rate": 4.2254409769335135e-05, "loss": 0.7662, "step": 18100 }, { "epoch": 23.42, "learning_rate": 4.2050881953867026e-05, "loss": 0.7779, "step": 18200 }, { "epoch": 23.55, "learning_rate": 4.184735413839891e-05, "loss": 0.7867, "step": 18300 }, { "epoch": 23.68, "learning_rate": 4.1643826322930797e-05, "loss": 0.7645, "step": 18400 }, { "epoch": 23.81, "learning_rate": 4.144029850746269e-05, "loss": 0.7815, "step": 18500 }, { "epoch": 23.81, "eval_loss": 0.17323149740695953, "eval_runtime": 656.7033, "eval_samples_per_second": 12.64, "eval_steps_per_second": 1.581, "eval_wer": 0.16304328966773662, "step": 18500 }, { "epoch": 23.94, "learning_rate": 4.1236770691994574e-05, "loss": 0.7783, "step": 18600 }, { "epoch": 24.07, "learning_rate": 4.103324287652645e-05, "loss": 0.79, "step": 18700 }, { "epoch": 24.2, "learning_rate": 4.082971506105834e-05, "loss": 0.7503, "step": 18800 }, { "epoch": 24.32, "learning_rate": 4.062618724559022e-05, "loss": 0.7941, "step": 18900 }, { "epoch": 24.45, "learning_rate": 4.0422659430122114e-05, "loss": 0.767, "step": 19000 }, { "epoch": 24.45, "eval_loss": 0.17244413495063782, "eval_runtime": 658.5684, "eval_samples_per_second": 12.605, "eval_steps_per_second": 1.576, "eval_wer": 0.16479569705907868, "step": 19000 }, { "epoch": 24.58, "learning_rate": 4.0219131614654e-05, "loss": 0.7828, "step": 19100 }, { "epoch": 24.71, "learning_rate": 4.0015603799185884e-05, "loss": 0.7594, "step": 19200 }, { "epoch": 24.84, "learning_rate": 3.9812075983717776e-05, "loss": 0.7649, "step": 19300 }, { "epoch": 24.97, "learning_rate": 3.960854816824966e-05, "loss": 0.7619, "step": 19400 }, { "epoch": 25.1, "learning_rate": 3.9405020352781546e-05, "loss": 0.7672, "step": 19500 }, { "epoch": 25.1, "eval_loss": 0.1706162840127945, "eval_runtime": 662.0343, "eval_samples_per_second": 12.539, "eval_steps_per_second": 1.568, "eval_wer": 0.15959052658974582, "step": 19500 }, { "epoch": 25.23, "learning_rate": 3.9201492537313424e-05, "loss": 0.7685, "step": 19600 }, { "epoch": 25.35, "learning_rate": 3.899796472184531e-05, "loss": 0.7739, "step": 19700 }, { "epoch": 25.48, "learning_rate": 3.87944369063772e-05, "loss": 0.7729, "step": 19800 }, { "epoch": 25.61, "learning_rate": 3.8590909090909086e-05, "loss": 0.7573, "step": 19900 }, { "epoch": 25.74, "learning_rate": 3.838738127544097e-05, "loss": 0.7691, "step": 20000 }, { "epoch": 25.74, "eval_loss": 0.17184105515480042, "eval_runtime": 658.3755, "eval_samples_per_second": 12.608, "eval_steps_per_second": 1.577, "eval_wer": 0.16177669818686563, "step": 20000 }, { "epoch": 25.87, "learning_rate": 3.818385345997286e-05, "loss": 0.7646, "step": 20100 }, { "epoch": 26.0, "learning_rate": 3.798032564450475e-05, "loss": 0.754, "step": 20200 }, { "epoch": 26.13, "learning_rate": 3.7776797829036634e-05, "loss": 0.7629, "step": 20300 }, { "epoch": 26.25, "learning_rate": 3.757327001356852e-05, "loss": 0.7601, "step": 20400 }, { "epoch": 26.38, "learning_rate": 3.7369742198100404e-05, "loss": 0.7547, "step": 20500 }, { "epoch": 26.38, "eval_loss": 0.1694486141204834, "eval_runtime": 662.0708, "eval_samples_per_second": 12.538, "eval_steps_per_second": 1.568, "eval_wer": 0.15648477487637721, "step": 20500 }, { "epoch": 26.51, "learning_rate": 3.716621438263229e-05, "loss": 0.7612, "step": 20600 }, { "epoch": 26.64, "learning_rate": 3.696268656716418e-05, "loss": 0.7699, "step": 20700 }, { "epoch": 26.77, "learning_rate": 3.675915875169606e-05, "loss": 0.7609, "step": 20800 }, { "epoch": 26.9, "learning_rate": 3.655563093622795e-05, "loss": 0.739, "step": 20900 }, { "epoch": 27.03, "learning_rate": 3.6352103120759836e-05, "loss": 0.7498, "step": 21000 }, { "epoch": 27.03, "eval_loss": 0.17055003345012665, "eval_runtime": 660.5207, "eval_samples_per_second": 12.567, "eval_steps_per_second": 1.571, "eval_wer": 0.15821983169948817, "step": 21000 }, { "epoch": 27.16, "learning_rate": 3.614857530529172e-05, "loss": 0.7583, "step": 21100 }, { "epoch": 27.28, "learning_rate": 3.5945047489823606e-05, "loss": 0.7591, "step": 21200 }, { "epoch": 27.41, "learning_rate": 3.574151967435549e-05, "loss": 0.7415, "step": 21300 }, { "epoch": 27.54, "learning_rate": 3.5537991858887376e-05, "loss": 0.734, "step": 21400 }, { "epoch": 27.67, "learning_rate": 3.533446404341927e-05, "loss": 0.7459, "step": 21500 }, { "epoch": 27.67, "eval_loss": 0.1662723422050476, "eval_runtime": 658.6276, "eval_samples_per_second": 12.603, "eval_steps_per_second": 1.576, "eval_wer": 0.1586362453370348, "step": 21500 }, { "epoch": 27.8, "learning_rate": 3.513093622795115e-05, "loss": 0.7579, "step": 21600 }, { "epoch": 27.93, "learning_rate": 3.492740841248304e-05, "loss": 0.7687, "step": 21700 }, { "epoch": 28.06, "learning_rate": 3.472388059701492e-05, "loss": 0.7493, "step": 21800 }, { "epoch": 28.19, "learning_rate": 3.452035278154681e-05, "loss": 0.7279, "step": 21900 }, { "epoch": 28.31, "learning_rate": 3.4316824966078694e-05, "loss": 0.7374, "step": 22000 }, { "epoch": 28.31, "eval_loss": 0.16514363884925842, "eval_runtime": 662.3435, "eval_samples_per_second": 12.533, "eval_steps_per_second": 1.567, "eval_wer": 0.15674503339984386, "step": 22000 }, { "epoch": 28.44, "learning_rate": 3.411329715061058e-05, "loss": 0.7525, "step": 22100 }, { "epoch": 28.57, "learning_rate": 3.3909769335142464e-05, "loss": 0.7336, "step": 22200 }, { "epoch": 28.7, "learning_rate": 3.370827679782903e-05, "loss": 0.7426, "step": 22300 }, { "epoch": 28.83, "learning_rate": 3.3504748982360924e-05, "loss": 0.7502, "step": 22400 }, { "epoch": 28.96, "learning_rate": 3.3303256445047485e-05, "loss": 0.7499, "step": 22500 }, { "epoch": 28.96, "eval_loss": 0.1668335199356079, "eval_runtime": 660.1378, "eval_samples_per_second": 12.575, "eval_steps_per_second": 1.572, "eval_wer": 0.1548538214626529, "step": 22500 }, { "epoch": 29.09, "learning_rate": 3.309972862957937e-05, "loss": 0.7375, "step": 22600 }, { "epoch": 29.21, "learning_rate": 3.289620081411126e-05, "loss": 0.7397, "step": 22700 }, { "epoch": 29.34, "learning_rate": 3.269267299864315e-05, "loss": 0.7362, "step": 22800 }, { "epoch": 29.47, "learning_rate": 3.248914518317503e-05, "loss": 0.7417, "step": 22900 }, { "epoch": 29.6, "learning_rate": 3.228561736770692e-05, "loss": 0.7471, "step": 23000 }, { "epoch": 29.6, "eval_loss": 0.16665612161159515, "eval_runtime": 665.4206, "eval_samples_per_second": 12.475, "eval_steps_per_second": 1.56, "eval_wer": 0.15528758566843065, "step": 23000 }, { "epoch": 29.73, "learning_rate": 3.20820895522388e-05, "loss": 0.7211, "step": 23100 }, { "epoch": 29.86, "learning_rate": 3.187856173677069e-05, "loss": 0.7264, "step": 23200 }, { "epoch": 29.99, "learning_rate": 3.167503392130258e-05, "loss": 0.7621, "step": 23300 }, { "epoch": 30.12, "learning_rate": 3.147150610583446e-05, "loss": 0.7484, "step": 23400 }, { "epoch": 30.24, "learning_rate": 3.126797829036635e-05, "loss": 0.7369, "step": 23500 }, { "epoch": 30.24, "eval_loss": 0.16586658358573914, "eval_runtime": 659.5988, "eval_samples_per_second": 12.585, "eval_steps_per_second": 1.574, "eval_wer": 0.1555651947601284, "step": 23500 }, { "epoch": 30.37, "learning_rate": 3.1064450474898235e-05, "loss": 0.7319, "step": 23600 }, { "epoch": 30.5, "learning_rate": 3.086092265943012e-05, "loss": 0.7514, "step": 23700 }, { "epoch": 30.63, "learning_rate": 3.0657394843962005e-05, "loss": 0.7341, "step": 23800 }, { "epoch": 30.76, "learning_rate": 3.045386702849389e-05, "loss": 0.716, "step": 23900 }, { "epoch": 30.89, "learning_rate": 3.025033921302578e-05, "loss": 0.7389, "step": 24000 }, { "epoch": 30.89, "eval_loss": 0.1668270081281662, "eval_runtime": 659.9227, "eval_samples_per_second": 12.579, "eval_steps_per_second": 1.573, "eval_wer": 0.1537954368005552, "step": 24000 }, { "epoch": 31.02, "learning_rate": 3.0046811397557663e-05, "loss": 0.7348, "step": 24100 }, { "epoch": 31.15, "learning_rate": 2.9843283582089552e-05, "loss": 0.7297, "step": 24200 }, { "epoch": 31.27, "learning_rate": 2.9639755766621434e-05, "loss": 0.7189, "step": 24300 }, { "epoch": 31.4, "learning_rate": 2.9436227951153322e-05, "loss": 0.7254, "step": 24400 }, { "epoch": 31.53, "learning_rate": 2.9232700135685207e-05, "loss": 0.7197, "step": 24500 }, { "epoch": 31.53, "eval_loss": 0.1687065213918686, "eval_runtime": 661.8428, "eval_samples_per_second": 12.542, "eval_steps_per_second": 1.568, "eval_wer": 0.1561204129435239, "step": 24500 }, { "epoch": 31.66, "learning_rate": 2.9029172320217096e-05, "loss": 0.7213, "step": 24600 }, { "epoch": 31.79, "learning_rate": 2.8825644504748977e-05, "loss": 0.7266, "step": 24700 }, { "epoch": 31.92, "learning_rate": 2.8622116689280866e-05, "loss": 0.7088, "step": 24800 }, { "epoch": 32.05, "learning_rate": 2.841858887381275e-05, "loss": 0.7324, "step": 24900 }, { "epoch": 32.17, "learning_rate": 2.821506105834464e-05, "loss": 0.71, "step": 25000 }, { "epoch": 32.17, "eval_loss": 0.16655509173870087, "eval_runtime": 666.1737, "eval_samples_per_second": 12.461, "eval_steps_per_second": 1.558, "eval_wer": 0.1516092652034354, "step": 25000 }, { "epoch": 32.3, "learning_rate": 2.8011533242876524e-05, "loss": 0.7224, "step": 25100 }, { "epoch": 32.43, "learning_rate": 2.780800542740841e-05, "loss": 0.7158, "step": 25200 }, { "epoch": 32.56, "learning_rate": 2.7604477611940295e-05, "loss": 0.7154, "step": 25300 }, { "epoch": 32.69, "learning_rate": 2.7402985074626863e-05, "loss": 0.6996, "step": 25400 }, { "epoch": 32.82, "learning_rate": 2.719945725915875e-05, "loss": 0.7199, "step": 25500 }, { "epoch": 32.82, "eval_loss": 0.16397617757320404, "eval_runtime": 665.0744, "eval_samples_per_second": 12.481, "eval_steps_per_second": 1.561, "eval_wer": 0.152337989069142, "step": 25500 }, { "epoch": 32.95, "learning_rate": 2.6995929443690633e-05, "loss": 0.7132, "step": 25600 }, { "epoch": 33.08, "learning_rate": 2.679240162822252e-05, "loss": 0.7134, "step": 25700 }, { "epoch": 33.2, "learning_rate": 2.6588873812754406e-05, "loss": 0.6905, "step": 25800 }, { "epoch": 33.33, "learning_rate": 2.6385345997286295e-05, "loss": 0.7102, "step": 25900 }, { "epoch": 33.46, "learning_rate": 2.618181818181818e-05, "loss": 0.7194, "step": 26000 }, { "epoch": 33.46, "eval_loss": 0.1659289002418518, "eval_runtime": 660.5596, "eval_samples_per_second": 12.567, "eval_steps_per_second": 1.571, "eval_wer": 0.15280645441138196, "step": 26000 }, { "epoch": 33.59, "learning_rate": 2.5978290366350065e-05, "loss": 0.7092, "step": 26100 }, { "epoch": 33.72, "learning_rate": 2.5776797829036633e-05, "loss": 0.7152, "step": 26200 }, { "epoch": 33.85, "learning_rate": 2.557327001356852e-05, "loss": 0.7224, "step": 26300 }, { "epoch": 33.98, "learning_rate": 2.5369742198100407e-05, "loss": 0.7078, "step": 26400 }, { "epoch": 34.11, "learning_rate": 2.516621438263229e-05, "loss": 0.6923, "step": 26500 }, { "epoch": 34.11, "eval_loss": 0.1662396788597107, "eval_runtime": 674.0602, "eval_samples_per_second": 12.315, "eval_steps_per_second": 1.54, "eval_wer": 0.1507070356554177, "step": 26500 }, { "epoch": 34.23, "learning_rate": 2.4962686567164177e-05, "loss": 0.7214, "step": 26600 }, { "epoch": 34.36, "learning_rate": 2.4759158751696062e-05, "loss": 0.7136, "step": 26700 }, { "epoch": 34.49, "learning_rate": 2.455563093622795e-05, "loss": 0.6958, "step": 26800 }, { "epoch": 34.62, "learning_rate": 2.4352103120759832e-05, "loss": 0.7271, "step": 26900 }, { "epoch": 34.75, "learning_rate": 2.414857530529172e-05, "loss": 0.7054, "step": 27000 }, { "epoch": 34.75, "eval_loss": 0.16411349177360535, "eval_runtime": 660.8038, "eval_samples_per_second": 12.562, "eval_steps_per_second": 1.571, "eval_wer": 0.14857291576299123, "step": 27000 }, { "epoch": 34.88, "learning_rate": 2.3945047489823606e-05, "loss": 0.6913, "step": 27100 }, { "epoch": 35.01, "learning_rate": 2.3741519674355494e-05, "loss": 0.697, "step": 27200 }, { "epoch": 35.14, "learning_rate": 2.3537991858887383e-05, "loss": 0.6981, "step": 27300 }, { "epoch": 35.26, "learning_rate": 2.3334464043419264e-05, "loss": 0.6934, "step": 27400 }, { "epoch": 35.39, "learning_rate": 2.313093622795115e-05, "loss": 0.6955, "step": 27500 }, { "epoch": 35.39, "eval_loss": 0.16339896619319916, "eval_runtime": 679.5402, "eval_samples_per_second": 12.216, "eval_steps_per_second": 1.528, "eval_wer": 0.14966600156155113, "step": 27500 }, { "epoch": 35.52, "learning_rate": 2.2927408412483038e-05, "loss": 0.7035, "step": 27600 }, { "epoch": 35.65, "learning_rate": 2.2723880597014926e-05, "loss": 0.6886, "step": 27700 }, { "epoch": 35.78, "learning_rate": 2.2522388059701488e-05, "loss": 0.7097, "step": 27800 }, { "epoch": 35.91, "learning_rate": 2.2318860244233376e-05, "loss": 0.6856, "step": 27900 }, { "epoch": 36.04, "learning_rate": 2.211533242876526e-05, "loss": 0.7084, "step": 28000 }, { "epoch": 36.04, "eval_loss": 0.16183346509933472, "eval_runtime": 676.475, "eval_samples_per_second": 12.271, "eval_steps_per_second": 1.534, "eval_wer": 0.1477574390561291, "step": 28000 }, { "epoch": 36.16, "learning_rate": 2.191180461329715e-05, "loss": 0.6912, "step": 28100 }, { "epoch": 36.29, "learning_rate": 2.170827679782903e-05, "loss": 0.6962, "step": 28200 }, { "epoch": 36.42, "learning_rate": 2.150474898236092e-05, "loss": 0.6828, "step": 28300 }, { "epoch": 36.55, "learning_rate": 2.1301221166892805e-05, "loss": 0.6844, "step": 28400 }, { "epoch": 36.68, "learning_rate": 2.1097693351424694e-05, "loss": 0.6917, "step": 28500 }, { "epoch": 36.68, "eval_loss": 0.1589200794696808, "eval_runtime": 681.9046, "eval_samples_per_second": 12.173, "eval_steps_per_second": 1.522, "eval_wer": 0.14708076689511582, "step": 28500 }, { "epoch": 36.81, "learning_rate": 2.0894165535956582e-05, "loss": 0.6813, "step": 28600 }, { "epoch": 36.94, "learning_rate": 2.0690637720488464e-05, "loss": 0.6917, "step": 28700 }, { "epoch": 37.07, "learning_rate": 2.048710990502035e-05, "loss": 0.6864, "step": 28800 }, { "epoch": 37.19, "learning_rate": 2.0283582089552237e-05, "loss": 0.6789, "step": 28900 }, { "epoch": 37.32, "learning_rate": 2.0080054274084126e-05, "loss": 0.687, "step": 29000 }, { "epoch": 37.32, "eval_loss": 0.15887242555618286, "eval_runtime": 671.8399, "eval_samples_per_second": 12.356, "eval_steps_per_second": 1.545, "eval_wer": 0.1450160492756138, "step": 29000 }, { "epoch": 37.45, "learning_rate": 1.9876526458616007e-05, "loss": 0.6909, "step": 29100 }, { "epoch": 37.58, "learning_rate": 1.9672998643147893e-05, "loss": 0.6757, "step": 29200 }, { "epoch": 37.71, "learning_rate": 1.946947082767978e-05, "loss": 0.6963, "step": 29300 }, { "epoch": 37.84, "learning_rate": 1.926594301221167e-05, "loss": 0.693, "step": 29400 }, { "epoch": 37.97, "learning_rate": 1.9062415196743555e-05, "loss": 0.6914, "step": 29500 }, { "epoch": 37.97, "eval_loss": 0.15876132249832153, "eval_runtime": 679.0742, "eval_samples_per_second": 12.224, "eval_steps_per_second": 1.529, "eval_wer": 0.14645614643879587, "step": 29500 }, { "epoch": 38.1, "learning_rate": 1.8858887381275436e-05, "loss": 0.6769, "step": 29600 }, { "epoch": 38.22, "learning_rate": 1.8655359565807325e-05, "loss": 0.675, "step": 29700 }, { "epoch": 38.35, "learning_rate": 1.8451831750339213e-05, "loss": 0.6785, "step": 29800 }, { "epoch": 38.48, "learning_rate": 1.82483039348711e-05, "loss": 0.6719, "step": 29900 }, { "epoch": 38.61, "learning_rate": 1.8044776119402983e-05, "loss": 0.6646, "step": 30000 }, { "epoch": 38.61, "eval_loss": 0.16017936170101166, "eval_runtime": 683.2054, "eval_samples_per_second": 12.15, "eval_steps_per_second": 1.519, "eval_wer": 0.14682050837164917, "step": 30000 }, { "epoch": 38.74, "learning_rate": 1.784124830393487e-05, "loss": 0.6905, "step": 30100 }, { "epoch": 38.87, "learning_rate": 1.7637720488466757e-05, "loss": 0.6724, "step": 30200 }, { "epoch": 39.0, "learning_rate": 1.7434192672998642e-05, "loss": 0.6501, "step": 30300 }, { "epoch": 39.12, "learning_rate": 1.7230664857530527e-05, "loss": 0.6705, "step": 30400 }, { "epoch": 39.25, "learning_rate": 1.7027137042062416e-05, "loss": 0.6667, "step": 30500 }, { "epoch": 39.25, "eval_loss": 0.1588352620601654, "eval_runtime": 689.4941, "eval_samples_per_second": 12.039, "eval_steps_per_second": 1.505, "eval_wer": 0.14435672768283161, "step": 30500 }, { "epoch": 39.38, "learning_rate": 1.68236092265943e-05, "loss": 0.6622, "step": 30600 }, { "epoch": 39.51, "learning_rate": 1.6620081411126186e-05, "loss": 0.6746, "step": 30700 }, { "epoch": 39.64, "learning_rate": 1.641655359565807e-05, "loss": 0.6753, "step": 30800 }, { "epoch": 39.77, "learning_rate": 1.621302578018996e-05, "loss": 0.6749, "step": 30900 }, { "epoch": 39.9, "learning_rate": 1.6011533242876524e-05, "loss": 0.6754, "step": 31000 }, { "epoch": 39.9, "eval_loss": 0.15873819589614868, "eval_runtime": 677.5943, "eval_samples_per_second": 12.251, "eval_steps_per_second": 1.532, "eval_wer": 0.14550186518608485, "step": 31000 }, { "epoch": 40.03, "learning_rate": 1.5808005427408413e-05, "loss": 0.6753, "step": 31100 }, { "epoch": 40.15, "learning_rate": 1.5604477611940298e-05, "loss": 0.6644, "step": 31200 }, { "epoch": 40.28, "learning_rate": 1.5400949796472183e-05, "loss": 0.6619, "step": 31300 }, { "epoch": 40.41, "learning_rate": 1.519742198100407e-05, "loss": 0.6597, "step": 31400 }, { "epoch": 40.54, "learning_rate": 1.4993894165535955e-05, "loss": 0.6632, "step": 31500 }, { "epoch": 40.54, "eval_loss": 0.15860143303871155, "eval_runtime": 682.8945, "eval_samples_per_second": 12.156, "eval_steps_per_second": 1.52, "eval_wer": 0.14609178450594257, "step": 31500 }, { "epoch": 40.67, "learning_rate": 1.4790366350067841e-05, "loss": 0.6584, "step": 31600 }, { "epoch": 40.8, "learning_rate": 1.4586838534599726e-05, "loss": 0.6524, "step": 31700 }, { "epoch": 40.93, "learning_rate": 1.4383310719131613e-05, "loss": 0.6502, "step": 31800 }, { "epoch": 41.06, "learning_rate": 1.4179782903663498e-05, "loss": 0.6773, "step": 31900 }, { "epoch": 41.18, "learning_rate": 1.3976255088195385e-05, "loss": 0.6619, "step": 32000 }, { "epoch": 41.18, "eval_loss": 0.15714022517204285, "eval_runtime": 688.8586, "eval_samples_per_second": 12.05, "eval_steps_per_second": 1.507, "eval_wer": 0.14406176802290274, "step": 32000 }, { "epoch": 41.31, "learning_rate": 1.377272727272727e-05, "loss": 0.645, "step": 32100 }, { "epoch": 41.44, "learning_rate": 1.3569199457259157e-05, "loss": 0.6433, "step": 32200 }, { "epoch": 41.57, "learning_rate": 1.3365671641791044e-05, "loss": 0.6376, "step": 32300 }, { "epoch": 41.7, "learning_rate": 1.3162143826322929e-05, "loss": 0.6542, "step": 32400 }, { "epoch": 41.83, "learning_rate": 1.2958616010854816e-05, "loss": 0.6561, "step": 32500 }, { "epoch": 41.83, "eval_loss": 0.15637381374835968, "eval_runtime": 690.1652, "eval_samples_per_second": 12.028, "eval_steps_per_second": 1.504, "eval_wer": 0.14204910210809404, "step": 32500 }, { "epoch": 41.96, "learning_rate": 1.27550881953867e-05, "loss": 0.6565, "step": 32600 }, { "epoch": 42.08, "learning_rate": 1.2551560379918588e-05, "loss": 0.6591, "step": 32700 }, { "epoch": 42.21, "learning_rate": 1.2348032564450473e-05, "loss": 0.6511, "step": 32800 }, { "epoch": 42.34, "learning_rate": 1.214450474898236e-05, "loss": 0.6532, "step": 32900 }, { "epoch": 42.47, "learning_rate": 1.1940976933514244e-05, "loss": 0.6492, "step": 33000 }, { "epoch": 42.47, "eval_loss": 0.15392744541168213, "eval_runtime": 693.1544, "eval_samples_per_second": 11.976, "eval_steps_per_second": 1.498, "eval_wer": 0.14373210722651167, "step": 33000 }, { "epoch": 42.6, "learning_rate": 1.1739484396200814e-05, "loss": 0.6624, "step": 33100 }, { "epoch": 42.73, "learning_rate": 1.1535956580732698e-05, "loss": 0.6608, "step": 33200 }, { "epoch": 42.86, "learning_rate": 1.1332428765264586e-05, "loss": 0.6556, "step": 33300 }, { "epoch": 42.99, "learning_rate": 1.112890094979647e-05, "loss": 0.6382, "step": 33400 }, { "epoch": 43.11, "learning_rate": 1.0925373134328358e-05, "loss": 0.6649, "step": 33500 }, { "epoch": 43.11, "eval_loss": 0.15117992460727692, "eval_runtime": 672.9964, "eval_samples_per_second": 12.334, "eval_steps_per_second": 1.542, "eval_wer": 0.14059165437668084, "step": 33500 }, { "epoch": 43.24, "learning_rate": 1.0721845318860245e-05, "loss": 0.631, "step": 33600 }, { "epoch": 43.37, "learning_rate": 1.051831750339213e-05, "loss": 0.6476, "step": 33700 }, { "epoch": 43.5, "learning_rate": 1.0314789687924017e-05, "loss": 0.6498, "step": 33800 }, { "epoch": 43.63, "learning_rate": 1.0111261872455902e-05, "loss": 0.6426, "step": 33900 }, { "epoch": 43.76, "learning_rate": 9.907734056987789e-06, "loss": 0.6511, "step": 34000 }, { "epoch": 43.76, "eval_loss": 0.1538574993610382, "eval_runtime": 664.8185, "eval_samples_per_second": 12.486, "eval_steps_per_second": 1.561, "eval_wer": 0.1383534310748677, "step": 34000 }, { "epoch": 43.89, "learning_rate": 9.704206241519674e-06, "loss": 0.6355, "step": 34100 }, { "epoch": 44.02, "learning_rate": 9.50067842605156e-06, "loss": 0.6512, "step": 34200 }, { "epoch": 44.14, "learning_rate": 9.297150610583445e-06, "loss": 0.6472, "step": 34300 }, { "epoch": 44.27, "learning_rate": 9.093622795115332e-06, "loss": 0.6421, "step": 34400 }, { "epoch": 44.4, "learning_rate": 8.890094979647217e-06, "loss": 0.6551, "step": 34500 }, { "epoch": 44.4, "eval_loss": 0.152027428150177, "eval_runtime": 659.3672, "eval_samples_per_second": 12.589, "eval_steps_per_second": 1.574, "eval_wer": 0.13844018391602325, "step": 34500 }, { "epoch": 44.53, "learning_rate": 8.686567164179104e-06, "loss": 0.6483, "step": 34600 }, { "epoch": 44.66, "learning_rate": 8.48303934871099e-06, "loss": 0.6577, "step": 34700 }, { "epoch": 44.79, "learning_rate": 8.279511533242876e-06, "loss": 0.6509, "step": 34800 }, { "epoch": 44.92, "learning_rate": 8.075983717774763e-06, "loss": 0.634, "step": 34900 }, { "epoch": 45.05, "learning_rate": 7.872455902306648e-06, "loss": 0.6452, "step": 35000 }, { "epoch": 45.05, "eval_loss": 0.1509818583726883, "eval_runtime": 663.5705, "eval_samples_per_second": 12.51, "eval_steps_per_second": 1.564, "eval_wer": 0.13679187993406783, "step": 35000 }, { "epoch": 45.17, "learning_rate": 7.668928086838535e-06, "loss": 0.6415, "step": 35100 }, { "epoch": 45.3, "learning_rate": 7.46540027137042e-06, "loss": 0.6385, "step": 35200 }, { "epoch": 45.43, "learning_rate": 7.261872455902306e-06, "loss": 0.6495, "step": 35300 }, { "epoch": 45.56, "learning_rate": 7.058344640434192e-06, "loss": 0.6348, "step": 35400 }, { "epoch": 45.69, "learning_rate": 6.8548168249660775e-06, "loss": 0.6155, "step": 35500 }, { "epoch": 45.69, "eval_loss": 0.15219098329544067, "eval_runtime": 665.3938, "eval_samples_per_second": 12.475, "eval_steps_per_second": 1.56, "eval_wer": 0.13750325323154333, "step": 35500 }, { "epoch": 45.82, "learning_rate": 6.6512890094979635e-06, "loss": 0.6502, "step": 35600 }, { "epoch": 45.95, "learning_rate": 6.447761194029849e-06, "loss": 0.6522, "step": 35700 }, { "epoch": 46.07, "learning_rate": 6.244233378561737e-06, "loss": 0.6241, "step": 35800 }, { "epoch": 46.2, "learning_rate": 6.040705563093623e-06, "loss": 0.6403, "step": 35900 }, { "epoch": 46.33, "learning_rate": 5.837177747625509e-06, "loss": 0.628, "step": 36000 }, { "epoch": 46.33, "eval_loss": 0.15221403539180756, "eval_runtime": 658.0569, "eval_samples_per_second": 12.614, "eval_steps_per_second": 1.577, "eval_wer": 0.13660102368352564, "step": 36000 }, { "epoch": 46.46, "learning_rate": 5.633649932157395e-06, "loss": 0.6394, "step": 36100 }, { "epoch": 46.59, "learning_rate": 5.430122116689281e-06, "loss": 0.6202, "step": 36200 }, { "epoch": 46.72, "learning_rate": 5.226594301221167e-06, "loss": 0.6261, "step": 36300 }, { "epoch": 46.85, "learning_rate": 5.023066485753053e-06, "loss": 0.6158, "step": 36400 }, { "epoch": 46.97, "learning_rate": 4.8195386702849386e-06, "loss": 0.6389, "step": 36500 }, { "epoch": 46.97, "eval_loss": 0.15134426951408386, "eval_runtime": 663.8499, "eval_samples_per_second": 12.504, "eval_steps_per_second": 1.564, "eval_wer": 0.13774616118677888, "step": 36500 }, { "epoch": 47.1, "learning_rate": 4.6160108548168245e-06, "loss": 0.6383, "step": 36600 }, { "epoch": 47.23, "learning_rate": 4.4124830393487104e-06, "loss": 0.638, "step": 36700 }, { "epoch": 47.36, "learning_rate": 4.208955223880596e-06, "loss": 0.6193, "step": 36800 }, { "epoch": 47.49, "learning_rate": 4.005427408412483e-06, "loss": 0.6313, "step": 36900 }, { "epoch": 47.62, "learning_rate": 3.801899592944369e-06, "loss": 0.6265, "step": 37000 }, { "epoch": 47.62, "eval_loss": 0.15123163163661957, "eval_runtime": 661.0461, "eval_samples_per_second": 12.557, "eval_steps_per_second": 1.57, "eval_wer": 0.13691333391168561, "step": 37000 }, { "epoch": 47.75, "learning_rate": 3.5983717774762546e-06, "loss": 0.6362, "step": 37100 }, { "epoch": 47.88, "learning_rate": 3.3948439620081405e-06, "loss": 0.6336, "step": 37200 }, { "epoch": 48.01, "learning_rate": 3.1913161465400265e-06, "loss": 0.6276, "step": 37300 }, { "epoch": 48.13, "learning_rate": 2.9877883310719132e-06, "loss": 0.623, "step": 37400 }, { "epoch": 48.26, "learning_rate": 2.784260515603799e-06, "loss": 0.6197, "step": 37500 }, { "epoch": 48.26, "eval_loss": 0.1511179655790329, "eval_runtime": 661.0804, "eval_samples_per_second": 12.557, "eval_steps_per_second": 1.57, "eval_wer": 0.136184610045979, "step": 37500 }, { "epoch": 48.39, "learning_rate": 2.580732700135685e-06, "loss": 0.6209, "step": 37600 }, { "epoch": 48.52, "learning_rate": 2.377204884667571e-06, "loss": 0.6361, "step": 37700 }, { "epoch": 48.65, "learning_rate": 2.173677069199457e-06, "loss": 0.6226, "step": 37800 }, { "epoch": 48.78, "learning_rate": 1.970149253731343e-06, "loss": 0.6446, "step": 37900 }, { "epoch": 48.91, "learning_rate": 1.7666214382632293e-06, "loss": 0.621, "step": 38000 }, { "epoch": 48.91, "eval_loss": 0.1510026901960373, "eval_runtime": 664.8917, "eval_samples_per_second": 12.485, "eval_steps_per_second": 1.561, "eval_wer": 0.13566409299904572, "step": 38000 }, { "epoch": 49.03, "learning_rate": 1.5630936227951152e-06, "loss": 0.6328, "step": 38100 }, { "epoch": 49.16, "learning_rate": 1.3595658073270013e-06, "loss": 0.6267, "step": 38200 }, { "epoch": 49.29, "learning_rate": 1.1560379918588873e-06, "loss": 0.6346, "step": 38300 }, { "epoch": 49.42, "learning_rate": 9.525101763907733e-07, "loss": 0.647, "step": 38400 }, { "epoch": 49.55, "learning_rate": 7.489823609226593e-07, "loss": 0.6259, "step": 38500 }, { "epoch": 49.55, "eval_loss": 0.15063393115997314, "eval_runtime": 657.6897, "eval_samples_per_second": 12.621, "eval_steps_per_second": 1.578, "eval_wer": 0.13528238049796132, "step": 38500 }, { "epoch": 49.68, "learning_rate": 5.454545454545454e-07, "loss": 0.632, "step": 38600 }, { "epoch": 49.81, "learning_rate": 3.419267299864315e-07, "loss": 0.6375, "step": 38700 }, { "epoch": 49.94, "learning_rate": 1.3839891451831749e-07, "loss": 0.6163, "step": 38800 }, { "epoch": 50.0, "step": 38850, "total_flos": 1.5592073449484252e+20, "train_loss": 0.9556344533304143, "train_runtime": 205033.5418, "train_samples_per_second": 6.067, "train_steps_per_second": 0.189 } ], "max_steps": 38850, "num_train_epochs": 50, "total_flos": 1.5592073449484252e+20, "trial_name": null, "trial_params": null }